Due to the potential for adverse selection, the cost of selling insurance depends not only on the number of policies sold but also to whom they are sold. This differentiates insurance markets from conventional markets and admits novel strategies (segmentation strategies) through which firms may exert market power. In the first chapter, I develop a two-stage spatial model of Bertrand price competition, with an endogenously determined rule for sharing demand, to examine the role that information asymmetries play between providers of insurance. These information asymmetries, which concern the expected cost of entering into an insurance contract, permit an advantaged firm to discriminate more precisely between consumers than its competitors. It follows that the advantaged firm can increase its competitors' costs by generating adverse selection problems for its competitors solely through its superior ability to discriminate and its pricing strategy. Furthermore, the advantaged firm is able to isolate subsets of the market from competition and price them above cost even as free entry and other competitive forces ensure marginal cost pricing throughout the rest of the market. The model explains anomalous features of insurance markets, including the failure of the law of one price and premium inflation. Comparative statics are used to analyze relevant regulatory issues and the welfare implications of these strategies are discussed. Finally, a broad measure of the ability of firms to discriminate between consumers is constructed to test implications of the model. The second chapter expands the empirical analysis of these strategies. Evidence is presented of price dispersion and strategic differentiation in categorization consistent with the use of segmentation strategies. Additional market irregularities are tested for using a data set constructed with consumer categorization and price data gathered from the Washington automobile insurance market. An instrumental variables approach is employed to identify the relationship between two measures of selection advantage and each of the following characteristics: loss results, market share, and rate level. In each case, the results are statistically significant and consistent with the effective use of segmentation strategies. Additional analysis supports the robustness of these results, including the use of alternative measures of selection advantage.
展开▼